summaryrefslogtreecommitdiff
path: root/build2/test
diff options
context:
space:
mode:
authorBoris Kolpackov <boris@codesynthesis.com>2016-09-25 03:43:37 +0200
committerBoris Kolpackov <boris@codesynthesis.com>2016-09-25 03:43:37 +0200
commit344c0a248c9cbec4c4ab844d903be6347589d8e9 (patch)
tree9ec839b69d688249cebe609bf0ad90b676413ee1 /build2/test
parent275854f9f153f52cf1672a5a18bf56fed91ddd56 (diff)
Add idea: Incremental tests: only run tests that were updated
Diffstat (limited to 'build2/test')
-rw-r--r--build2/test/list10
1 files changed, 10 insertions, 0 deletions
diff --git a/build2/test/list b/build2/test/list
new file mode 100644
index 0000000..58b23c7
--- /dev/null
+++ b/build2/test/list
@@ -0,0 +1,10 @@
+- Incremental tests: only run tests that were updated [idea]
+
+ If a test is more than an executable, then tracking its "inputs" (so that we
+ know whether to run it) will be tricky. Such a test will most likely run via
+ testscript so can we capture inputs somehow? This seems to be related to the
+ question of what to attach testscripts to. Maybe we should make them (in
+ source) targets?
+
+ Will also need some kind of switch/override to run all tests (command line
+ config override seems like a natural choice).